Astm stp 518 1983

KEY WORDS: stress corrosion cracking, corrosion, fracture properties, fracture tests, corrosion tests, crack propagation, pitting, notch tests Stress corrosion cracking SCC is one of t

Trang 2

STRESS CORROSION

CRACKING OF

METALS-A STMETALS-ATE OF THE METALS-ART

A symposium presented at the American Society for Metals Metals Conference

Detroit, Michigan, 18 October 1971

ASTM SPECIAL TECHNICAL PUBLICATION 518

H Lee Craig, Jr., symposium chairman

AlWlVEKSARy

04-518000-27

^ AMERICAN SOCIETY FOR TESTING A N D MATERIALS

1916 Race Street, Philadelphia, Pa 19103

Trang 3

Library of Congress Catalog Card Number: 72-85692

NOTE The Society is not responsible, as a body, for the statements and opinions advanced in this publication

Trang 4

FOREWORD

The Symposium on Stress Corrosion was presented at the American Society

for Metals, Metals Conference held in Detroit, Michigan, 18 October 1971

Subcommittee 6 on Stress Corrosion Cracking and Corrosion Fatigue of the

ASTM Committee G-1 on Corrosion of Metals sponsored the symposium H

Lee Craig, Jr., Reynolds Metals, Co., presided as symposium chairman

Trang 6

CONTENTS

Introduction 1

A Preface to the Problem of Stress Corrosion Cracking-B.F BROWN 3

Stress Corrosion Cracking of a High Strength Steel-A.M SHEINKER

AND J.D WOOD 16

Stress Corrosion Cracking of Copper Metals-D.H THOMPSON 39

Stress Corrosion Cracking Behavior of Nickel and Nickel AUoys-W.K

BOYD AND W.E BERRY 58

Testing Methods for Stress Corrosion Cracking-S.J KETCHAM 79

The Resistance of Wrought High Strength Aluminum Alloys to Stress

Corrosion Cracking-'D.O , SPROWLS, R.H BROWN, AND

M.B SHUMAKER 87

Overview of Corrosion Cracking of Titanium Alloys—N.G FEIGE AND

L.C.COVINGTON 119

Stress Corrosion Crack Protection from Coatings on High Strength

H-11 Steel Aerospace Bolts-EDWARD TAYLOR 131

Corrosion Fatigue at High Frequencies and Hydrostatic Pressures—A

THIRUVENGADAM 139

Resistance of High Strength Structural Steel to Environmental Stress

Corrosion-H.E TOWNSEND, JR 155

Trang 7

STP518-EB/Sep 1972

INTRODUCTION

This publication is a concrete example of the cooperation that exists

between technical societies - in this instance, the American Society for Metals

and the American Society for Testing and Materials Subcommittee 6 of

ASTM Committee G-1 on Corrosion of Metals presented a symposium on

stress corrosion at the Fall 1971 meeting of the ASM in Detroit, Michigan

These papers are based on the talks given at that time The objective was to

present a timely report on the state of stress corrosion from a practical,

engineering standpoint

The excellent attendance at this symposium was mute testimony to the

widespread nature of the problem of stress corrosion cracking Project

managers, designers, metallurgists, metallurgical engineers, each is concerned

with this problem Unexpected failure of metal parts has plagued the defense,

chemical, petroleum, and other industries However, analysis of each stress

corrosion failure is seldom surprising — usually one or more caveats have been

violated through ignorance, accident, or lack of precaution Many of us, active

in the field of stress corrosion, have come to the conclusion that the

educational part of our work is the most significant, from the standpoint of

prevention of failures

Thus, experts from all phases of the metals industry, from government

laboratories, research institutes and from universities gathered together to

present the best, current thinking about the problems and the solutions to the

use of high strength materials which may be susceptible to stress corrosion

cracking

In this volume information will be found on steels, including the new, high

strength steels as well as the stainless and mild steels Aluminum alloys are

discussed with emphasis on the newer versions of high strength alloys and

tempers specifically designed for stress corrosion resistance Other engineering

metals and their alloys are covered, including copper, titanium, and nickel

These materials are discussed in relation to the newer testing methods that

have evolved during the past decade Several authors develop the concepts of

linear elastic fracture mechanics as they are apphed to specimen design and

the interpretation of data However, the older, time tested methods are not

overlooked, as one author details the efforts of ASTM to standardize

specimens and solutions used in stress corrosion testing

This volume is presented to increase the understanding of the interested

person who has a need to deal with stress corrosion cracking, either in the

design of structures, the selection of materials, the specification of fabrication

or maintenance procedures, or regretfully, in failure analysis Each author was

encouraged to deal with his subject using a practical, engineering approach In

1

Copyright' 1972 by A S T M International www.astm.org

Downloaded/printed by

University of Washington (University of Washington) pursuant to License Agreement No further reproductions authorized.

Trang 8

addition, I encourage anyone who has an interest or a problem dealing with

stress corrosion, to become affiliated with Subcommittee 6 of Committee G-1

on Stress Corrosion Cracking and Corrosion Fatigue, and work with us in the

development of standard test methods that will be used to help select

materials and thereby minimize failures from stress corrosion cracking

H Lee Craig, Jr

Reynolds Metals Company Metallurgical Research Division 4th and Canal Streets

Richmond, Va 23261 symposium chairman

Trang 9

B F Brown i

A Preface to the Problem of

Stress Corrosion Cracking

REFERENCE: Brown, B.F., "A Preface to the Problem of Stress Corrosion

Cracking," Stress Corrosion Cracking of Metals-A State of the Art, ASTM STP

518, American Society for Testing and Materials, 1972, pp 3-15

ABSTRACT: The characteristics of stress corrosion cracking (SCC) are enumerated

in the context of a historical sketch of the problem The roles of pitting and

brittle fracture in affecting the behavior of materials in tests of smooth specimens

are depicted The rationale for using fracture mechanics in evaluating crack

propagation behavior is given, and a rudimentary composite of the results of smooth

specimen tests and crack propagation ("fracture mechanics") specimen tests is

presented We lack predictive capability with respect to SCC from one chemical

environment to another

KEY WORDS: stress corrosion cracking, corrosion, fracture properties, fracture

tests, corrosion tests, crack propagation, pitting, notch tests

Stress corrosion cracking (SCC) is one of those irritating considerations of

the designer who may have to select materials of construction to meet a series

of other property requirements that cannot be waived at all or can only be

waived within narrow limits The stress corrosion problem must therefore be

considered in the context of the other constraints on design and maintenance,

including costs The designer who must use high strength materials will not be

able to select structural alloys which are totally immune to SCC, so that he

must understand the meaning of test data characterizing susceptibility to this

form of failure The alloy developer also needs to understand the significance

of macroscopic characterization data since theory is inadequate to guide alloy

development

Much of this introductory paper will therefore treat macroscopic

phenome-na, macroscopic tests, and the interpretation of macroscopic test data It is

helpful first to summarize the characteristics of SCC, which is conveniently

done in a historical review of the problem

'Metallurgy Division, Naval Research Laboratory, Washington, D.C

Copyright' 1972 by AS FM International www.astm.org

Trang 10

4 STRESS CORROSION CRACKING OF METALS

Historical Sketch

s e c first became a widespread problem with the introduction of the cold

drawn brass cartridge case during the last half of the 19th century Toward

the end of the century it appeared in the brass (but not in the unalloyed

copper) condenser tubing in the rapidly growing electric power generation

industry During this era the problem became sufficiently important to acquire

its own name, "season cracking." Also during this period Professor W

Chandler Roberts-Austen (whence "austenite") made two important

contribu-tions to the problem: he showed that a cold drawn wire of an alloy of gold,

silver, and copper would undergo SCC if touched with a drop of ferric

chloride solution, thus demonstrating that the phenomenon was not restricted

to brass or even to base alloys And in analyzing the stresses in the wire he

placed emphasis for the first time on the necessary role of tensile stress in

SCC By the close of the 19th century the role of residual stresses in causing

SCC in brass was so widely known that stress relieving heat treatments for

cold formed products were developed as mitigative measures, and acidified

mercurous nitrate, which will cause mercury cracking ("liquid metal

embrittle-ment") of stressed brass, was in widespread use to verify the effectiveness of a

stress relief treatment for a given brass product

It will be appreciated that the tensile stresses which caused SCC in

cartridge cases and in condenser tubes were residual stresses caused by cold

forming operations and that therfore the stress fields tended to have complex

geometry As a consequence, it is not surprising that the stress corrosion

cracks seen in this era tended to branch in response to the geometric

complications of the stress fields, and that this branching was so common that

it has remained accepted as a characteristic of SCC We will see that there is

another reason for branching of stress corrosion cracks But under many

practical circumstances where SCC is caused by service stresses, branching may

be entirely absent

It was during the late 19th century that ammonia was found to play a

causative role in the SCC of brass, a finding which contributed to the

development of the rule that there is a specificity of environment which will

cause SCC in a given alloy This specificity is usually cited as a prime

characteristic of SCC, but the growing number of known exceptions makes

the specificity rule of questionable merit Regardless of the question of

specificity, it became obvious that the responsible species need not be present

either in large quantities or in high concentration in most cases, at least not in

high concentration in the bulk environment

By the end of the 19th century "caustic cracking" of riveted boiler steel

could also be cited as an example of SCC and as another indication that the

problem might occur in a number of alloy systems, given the wrong

conditions The wrong condiUons in caustic cracking are a combination of

local high concentration of free alkaH and elevated temperature

Trang 11

BROWN ON A PREFACE TO THE PROBLEM 5

By this point the pattern of failures and mitigative measures had become

discernible: there are three elements of the phenomenon, mechanics,

metal-lurgy, and chemistry of the environment A given problem can be solved only

by changing one or more of these three elements, and valid research in the

laboratory requires adequate attention to all three

Early in the 20th century SCC was seen in aluminum alloys, attributable to

atmospheric moisture Also during this period SCC was observed in martensitic

steel, but the problem did not become widely recognized for what it was until

the era of the aerospace programs

Also early in the 20th century the cracking of mild steel due to nitrates

became of practical importance in the chemical industry From this experience

we have a clear statement of an important characteristic of SCC, namely, that

often it occurs when the alloy is almost inert to the environment which does

the cracking Based on experience with evaporating sodium nitrate and sodium

chloride solutions in steel pans, Professor Porter pointed out that

the action upon the steel if totally different in the case of the two

solutions You may go to a waste heap and pick out the pieces of steel

that have come from a sodium nitrate pan Those that have come from

a sodium chloride pan are all rusty, the steel rusted through, while those

from the sodium nitrate pan are not rusted at all, but they are cracked

[1]

During the 1930s when stainless steels came to be used extensively in the

paper, chemical, and petroleum industries, SCC was observed in this class of

alloys, particularly in chloride or caustic solutions at elevated temperature

Also during the 1930s, magnesium alloys for military aircraft were found to

be susceptible to SCC in moist atmospheres

During the 1950s the aerospace programs encountered, in addition to SCC

of martensitic steels indicated earlier, cracking of titanium alloys in contact

with nitric acid, or in contact with salt at high temperatures

During the 1960s, titanium alloys were also observed to be susceptible to

SCC in nitrogen tetroxide, in water, and in methanol Also, a zirconium alloy

was found to be susceptible to SCC in iodine bearing environments This

experience confirmed the growing supposition that SCC is a general

phenom-enon to be expected in all alloy systems

By 1960, the technique of producing high resolution rephcas of fracture

surfaces had been developed in basic form, and during the 1960s passed into

the hands of a large number of investigators Its application to SCC was

inevitable, and fruitful as well, for it disclosed that on the fracture surfaces

there were micron scale details important to the development of theory as

well as helpful in critical failure analyses This electron fractography

demon-strated the gross differences in the fracture topology on the micron scale

between stress corrosion cracks, which are always macroscopically brittle in

appearance, and "brittle fracture" in the same material The fact that stress

corrosion cracks are always macroscopically brittle, even in alloys which are

Trang 12

highly ductile in purely mechanical fracture toughness tests, is another general

characteristic of SCC

By the 1960s, fracture mechanics had matured to the point of being

applicable to treat the stress field around a stress corrosion crack in a useful

way, particularly after the analysis for the bent beam specimen made fracture

mechanics not only appHcable but also practical for long term tests In addition

to the fracture mechanics analysis, much of the methodology and some of the

insturmentation developed in fracture mechanics have been useful in the

evolution of SCC testing concepts and procedures

Characteristics of SCC

The characteristics of SCC may be summarized as follows:

1 Tensile stress is required This stress may be supphed by service loads,

cold work, mismatch in fit-up, heat treatment, and by the wedging action of

corrosion products

2 Only alloys are susceptible, not pure metals, though there may be a few

exceptions to this rule

3 Generally only a few chemical species in the environment are effective

in causing SCC of a given alloy

4 The species responsible for SCC in general need not be present either in

large quantities or in high concentrations

5 With some alloy/corrodent combinations, such as titanium and crystalline

sodium chloride, or austenitic stainless steel and chloride solutions,

tempera-tures substantially above room temperature may be required to activate some

process essential to SCC

6 An alloy is usually almost inert to the environment which causes SCC

7 Stress corrosion cracks are always macroscopically brittle in appearance,

even in alloys which are very tough in purely mechanical fracture tests (Shear

lips may occur in conjunction with stress corrosion cracks, but these shear lips

are not part of the stress corrosion process As a corollary, there does not

appear to be a stress corrosion fracture analog to the full shear slant fracture

of purely mechanical origin.)

8 Microscopically the fracture mode in SCC is usually different from the

fracture mode in plane strain fractures of the same alloy

9 There appears to be a threshold stress below which SCC does not occur,

at least in some systems This characteristic was not mentioned in the

foregoing historical sketch but will be treated in a subsequent section

Sequence of Events in SCC

In the most general case, if a smooth specimen is placed in a corrodent in

which it will eventually undergo SCC, the sequence of events is as shown in

the top row (Row A) of Fig 1 First a corrosion pit forms There is an

important feature of most corrosion pits, the significance of which is not

Trang 13

FIG l-Sequence of events (left to right) in alloys under stress in

a corrosive environment Materials in Rows A and B pit, and that

in Row B is the brittler Material of Rows C and D does not pit,

but specimen in Row D has a preexisting flaw

always appreciated, namely, a porous cap of corrosion products which must

be removed in order to see the pit itself This cap impedes exchange between

the corrodent within the pit and the bulk corrodent outside the pit, but it

permits inward migration of anions such as chloride This inward migration,

also seen during the growth of stress corrosion cracks, represents "uphill

diffusion" in terms of concentration of anions The driving force for this

"uphill diffusion" is of course the active metal surface In general the pH

within corrosion pits also differs from that outside the pit The function of a

corrosion pit in initiating SCC was once widely thought to be purely

mechanical, to concentrate stresses It now appears that the essential function

of the pit when it initiates SCC is not primarily mechanical, but is rather to

provide a mechanism for altering the solution chemistry locally to one

favorable for SCC

Continuing to the right in Row A of Fig 1, representing the passage of

time, we see in the third column a stress corrosion crack emanating from the

corrosion pit Assuming the stress is maintained, eventually the stress

corro-sion crack grows to such a length that the remaining metal ligament snaps in

purely mechanical brittle fracture (fourth column) The combination of stress

corrosion crack length and stress required to set off brittle fracture depends

upon the fracture toughness of the alloy, and in fact that combination is a

quantitative measure of the toughness Lower strength alloys such as brass,

austenitic stainless steels, and the older aluminum alloys are so tough that

brittle fracture would not occur in the usual laboratory specimens The

material in Row B is more brittle that that of Row A, and after a short stress

Trang 14

corrosion crack grows (third column), the remaining Ugament snaps Some

alloys are so brittle that the corrosion pit itself has been observed to initiate

not s e c but brittle fracture, so that although there is a "delayed fracturing,"

there is no SCC or other slow crack growth process

The material of Row C does not pit in the corrodent, and no SCC occurs

However, the same material containing a cracklike flaw at the surface, as in

Row D, may experience rapid SCC Titanium alloys in seawater are examples

of materials which behave as depicted in Rows C and D

It should be emphasized that neither a preexisting flaw nor a corrosion pit

is necessary for initiating SCC if the environment has the correct composition

for the alloỵ For example, although a titanium alloy may not undergo SCC in

salt water except from a preexisting crack or flaw, in methanol SCC initiates

readily at a smooth surface of the same alloỵ Also the precipitation hardening

steel designated "13-8 Mo" will not initiate SCC in neutral salt water under

given stress conditions until a corrosion pit is formed But if the salt water is

acidified with HCl, presumably to simulate the acidity within corrosion pits in

that steel, SCC initiates readily at a smooth surfacẹ

The important point of the foregoing discussion is that if one wishes to

estabhsh stress corrosion characteristics of a material, he must exclude possible

confusion from either pitting or nonpitting behavior on the one hand and

brittle fracture on the other

Mechanics

The budgets available to the corrosionist working on the SCC problem have

traditionally tended to encourage economy of experimentation Accordingly

simple strip specimens, stressed by bending in a simple fixture of one kind or

another, were typical Sometimes the specimens were notched, but not

precracked The characterization descriptors in such a test setup are

customar-ily the initial stress on the specimen and the time for failurẹ It is thought by

some that there is a limiting stress, designated ậ^ below which SCC does not

initiatẹ It is difficult if not impossible to establish the existence of a Oj-^ for

it would require proving a negativẹ If for example one demonstrates that no

SCC occurs below a specified stress level within a stated observation time, he

has not thereby excluded the posibihty that after a longer period SCC would

occur There does not seem to be any way around this difficultỵ

Neverthe-less, arbitrary though it may be, a defined threshold stress would seem to be a

useful thing to know

Much progress has been made in alloy technology by the use of such tests,

and also progress in the fundamentals of the process Inevitably however, this

test concept encountered difficulties when undetected effects of pitting

behavior and of fracture toughness caused serious experimental artifacts, as

discussed above in connection with Fig 1 It finally became clear that to

grapple successfully with the SCC problem one must study the cracking

process itself, especially its kinetics But the kinetics might reasonably be

Trang 15

assumed to be stress-dependent, so that any meaningful treatment of the

cracking process required a method to quantify stress in the presence of a

crack

Stress intensification theory of notches cannot be applied to an essentially

infinitely sharp crack because notch theory involves the reciprocal of the

notch root radius; in the case of the infinitely sharp crack the reciprocal

would he infinite and meaningless "Nominal" stress has been used extensively

to treat stress in a solid containing a crack by imagining removal of all the

cracked material A glance at Fig 2 will suggest intuitively the weakness of

FIG 2-"Nominal stress" treats a deeply cracked specimen as being equivalent to a smooth specimen equal in depth to the uncracked ligament of the cracked specimen

this procedure On the left in this figure a specimen is shown under stress

with a crack, and on the right a smooth specimen of the same thickness as

the unbroken ligament of the cracked specimen The nominal stress

rationah-zation would say that the stress situation at the bottom of the crack is the

same as on the surface of the uncracked specimen Not only is this

rationalization intuitively bothersome, but it has actually been demonstrated

to be fallacious [2] In short, a nominal stress in a cracked solid is a fiction

which can cause confusion and error

There is available, however, another way to quantify the stress factor in an

elastic body containing a crack, an analysis known as hnear elastic fracture

mechanics, or more commonly simply fracture mechanics The useful metals

and alloys are not purely elastic, particularly in the region around the tip of a

growing crack Thus fracture mechanics in its present form does not treat the

very porfion of the metal which is of most direct importance to SCC

Nevertheless, fracture mechanics can quantify the elastic stress field associated

with the crack, and it is this elastic field which produces the plastic zone at

the crack tip Since it is the elastic field which produces the plastic zone, we

can usefully apply elastic fracture mechanics even though it doesn't tell us

anything at all directly about the material at the crack tip Fracture mechanics

thus does not of itself give us new knowledge about SCC processes, it simply

provides a means of referencing stress in a body containing a crack in a

manner applicable to various geometries (It should be noted that in stating

that the elastic conditions control the plastic conditions, there may be an

important time dependence of this plasticity in some metals.)

Trang 16

10 STRESS CORROSION CRACKING OF METALS

The fracture mechanics descriptor of stress is useful for describing SCC

kinetics involving a lengthening crack, and it is also useful in using experience

gained with small specimens with one type of loading to predict the behavior

of larger components or structures of a different shape and with different

loading conditions For example, SCC behavior in a small bar stressed in

bending can be used, with fracture mechanics methodology, to predict the

behavior of large plates in tension

The fracture mechanics descriptor which has been most useful to date is

the stress intensity factor K, which is proportional to the stress multiplied by

a function of the square root of the crack length The usual units in the

United States are ksi/m

Studies of SCC kinetics as a function of K have shown that the crack

growth rate is often approximately an exponential function of /C up to a

certain value of K Beyond this K level, the rate is insensitive to K, as shown

in Fig 3 (after Ref 5) At still higher K levels there may be another

CRACK GROWTH RATE (LOG SCALE)

>

STRESS INTENSITY K

Iscc

FIG 3-Generalized representation of effect of K on SCC

kinetics There is a change in the scale of the ordinate so that the origin represents zero crack growth rate Some systems exhibit the abrupt drop in kinetics apparently to this zero rate, at a finite value of K, denoted by K/j^,^

After Ref 4

A^-dependent regime The features of Fig 3 emphasize the hazard of

con-ducting basic research studies involving cracking kinetics without knowing

which regime one is working in, and without knowing whether he is crossing

regime boundaries In the /T-independent regime there is a tendency for crack

branching caused by mechanics alone

Figure 3 also shows that in some systems the crack growth rate may drop

abruptly to zero, as far as can be detected The K level below which crack

growth has not been observed but above which it is observed (for a specified

combination of alloy and corrodent) has been termed A^ijcc- The usefulness of

such a parameter, for those systems for which it exists, is that it quantifies

Trang 17

s e c resistance with a single number, and a number which has predictive

capabilities with respect to combinations.of preexisting crack depths and stress

levels which would cause SCC If one assumes that any cracklike flaw in the

surface is long compared to its depth, and if he further assumes that yield

point stresses exist around this flaw, then if he knows the value of Ki^^^ for

this material in a given environment, he can estimate how deep the flaw must

be before it will initiate SCC by the equation

where a is the critical flaw depth and Oy is the yield strength

This method of characterizing SCC can be extended to provide a

conveni-ent method for displaying and comparing SCC characteristics of various alloys,

as shown in Fig 4 In this figure ATj^^p data for the alloy Ti-6A14V in salt

water is shown as a function of yield strength The lines drawn represent Eq 1

for four selected values of flaw depth The significance of these lines is as

follows: if a surface flaw can be present as deep as the value of a shown for a

given line, and if yield point stresses are present, one would want to use only

materials having ^jscc values above that line

FIG 4'-K/5^^ data for various heats of Ti-6AI-4V alloy in salt water,

illustrating one method for displaying the SCC resistance of various alloys for materials selection purposes Tfiree alloys would experience SCC if stressed at their yield strength if there were a long surface crack 0.1 in deep: one alloy would fail if the surface crack were only about 0.005 in deep Encircled area contains all other known data on K/^^.j,

of Ti-6A1 •4 V alloy in salt water

Trang 18

Those who have heard the rule that "fracture mechanics does not apply to

low strength materials" should be reminded that the limitations on fracture

mechanics are not fixed by strength alone but by the ratio of K to strength If

s e c occurs at sufficiently low values, then fracture mechanics is applicable

even to low strength materials as long as crack branching does not occur

No part of the SCC problem has been more troublesome than combining

the methodology of crack propagation using fracture mechanics and the

initiation and growth experience using initially smooth specimens If there is a

true SCC cracking threshold, it is designated ^jscc- '^ there is a threshold

stress for SCC initiating from an originally smooth surface, it is designated

Oj-ff The predictions of these two measures of SCC resistance may be

combined as in Fig 5 The curve represents the Irwin equation for a long

surface crack:

1^2 1.2 iro^a

(2) Iscc

1 0.2 ( ~ )

Oy

The horizontal line indicates the Oy-// In this figure one would expect SCC if

the stress is above the solid line or if the combination of stress and flaw size

lies to the right or above the curve

FIG 5-Combining SCC data from smooth specimens (ojy^)

and crack propagation ("fracture mechanics") specimens (the curve) If the data are valid, the intersection of the horizontal line and the curve should indicate a corrosion pit thai would reasonably function mechanically about the same

as a crack of the indicated depth at the intersection After Ref4

Trang 19

Unfortunately we do not have many data from which to construct a

diagram such as the schematic one of Fig 5 There has been one attempt to

make such a diagram for the aluminum alloy 7075-T6510 [4] In this instance

the intersection of the horizontal line and the curve occurred at a very large

value of a, as though a corrosion pit were acting like a very deep crack One

possible explanation for this behavior is that the alloy does not appear to

possess a real ATjj^p and that therefore cracking occurs at much lower K values

than that for which the curve was drawn (K^^^^ does not apply, and the

number assumed for it is too large) A second possible reason for the

unreasonably large equivalent value of the corrosion pit is that possibly the

solution within the corrosion pit may differ in an important way from the

solution near the tip of a growing crack Conceivably an important difference

in solution chemistry could make an important difference in K^^^^ and hence

in the position of the curve of Fig 5 and the intersection point

Local Solution Chemistry

The alterations which occur in the corroding solution within the cavity of a

pit while it is growing have already been mentioned It has been found that

the corrodent within stress corrosion cracks in high strength steels, aluminum

alloys, and titanium alloys is acid, as shown in Fig 6, even though the bulk

corrodent was nearly neutral (pH 6, due to dissolved carbon dioxide) This

acidification may be described in terms of traditional hydrolysis reactions,

such as a metal salt reacting with water Alternatively one can describe the

reactions as involving the combining of metal with oxide or hydroxyl ions,

leaving an excess of hydrogen ions These hydrogen ions are balanced by the

- 6

Al ALLOYS ( A I + ? )

T i ALLOYS ( T i * - ^ )

FIG 6-Potential and pH data at the tips of growing stress corrosion cracks in four

alloy families The bulk corrodent is nearly neutral salt water The broken line

indicates the Nernst potential for the reversible hydrogen electrode

Trang 20

influx of anions such as chloride The result is an acid solution which

simultaneously may promote the reduction of hydrogen and oppose the

repassivation of the metal surface In the instance of magnesium alloys, the

local pH is buffered to a high value by the presence of magnesium hydroxide

The acidic nature of the corrodent within stress corrosion cracks in the

heavy metals is only one reason that effective inhibitors for SCC propagation

are virtually nonexistent It is difficult enough to solve the problem of

inhibiting acid chloride solutions This task is made even more difficult by the

exceedingly small amount of corrodent within stress corrosion cracks, the

amount being too small to provide a sufficient reservoir for inhibitors For the

same reason, the addition of buffers to the bulk environment makes only a

modest change in stress corrosion crack propagation — the buffering capacity

of even concentrated buffers is not sufficient to overcome entirely the

hydrolytic acidification capacity of the clean metal surfaces of the crack

Mechanisms

Many investigators and commentators have asked the question whether

there is a single mechanism which can explain SCC in all the systems in which

it has been observed The prevailing present opinion is that such a common

explanation is improbable The mechanisms that have been advanced fall into

the following categories:

1 Mechano—electrochemical This category includes such models as the

high strain rate at the tip of the growing crack causing continuous

depolariza-tion of the anodic area so that cracking by electrochemical dissoludepolariza-tion can

continue As an added feature, another model envisions a separate role of

strain in causing the rupturing of islands of material resistant to

electro-chemical dissolution which has previously produced all the fracture area

except the resistant islands

2 Film rupture This model envisions the cyclic formation and rupture of

films of corrosion product

3 Embrittlement This model envisions electrochemical corrosion causing

embrittlement of the metal immediately behind the corroding surface This

model may picture either a continuous or a cyclic process

4 Adsorption This model envisions the reduction of energy to form a new

surface by adsorption of specific species

The interested reader may wish to study Refs 5 through 7 for more

detailed discussions of mechanisms

There is limited agreement among various investigators on the applicability

of any one of the above categories of mechanisms for a given combination of

metal and corrodent Furthermore, none of the models has been useful either

in alloy development or in predicting SCC in fundamentally new combinations

of alloys and environments

Trang 21

BROWN ON A PREFACE TO THE PROBLEM 15 Fracture mechanics has given us predictive capabihty from one geometry to

another, within hmits We are lacking predictive capabihties from one

environ-ment to another, and the outlook does not seem favorable for early substantial

improvement in this element of the problem Macroscopic test data will

therefore continue to be of paramount importance until theory is much better

in hand This is in no way intended as a criticism of the work on theory past

or present, merely a recognition of the status of this difficult problem

[2] Novak, S.R and Rolfe, S.T., Corrosion, Vol 26, 1970, p 121

[S] Wei, R.P et al in NRL Memorandum Report 1941, Naval Research Laboratory,

Washington, D.C., Oct 1968

[4] Hyatt, M.V., Document D6-24466, The Boeing Co., Seattle, Wash., Nov 1969; and

Hyatt, M.V and Schimmelbusch, H.W., AFML-TR-70-109, Air Force Materials

Laboratory, Dayton, Ohio, May 1970

[5] Scully, J.C., Ed., "The Theory of Stress Corrosion Cracking in Alloys," Proceedings

of a 1971 NATO Science Committe Conference, in press

[6] Staehle, R.W., Forty, A.J., and van Rooyen, D., Eds., "Fundamental Aspects of

Stress Corrosion Cracking," National Association of Corrosion Engineers, Houston,

Tex., 1969

[7] Pugh, E.N., Green, JA.S., and Sedriks, A.J., "Interfaces Conference - Melbourne

1969," Butterworths, 1969, p 237

Related References

Brown, BF., Metals and Materials, Vol 2, 1968, p 171

Wei, R.P., "Fundamental Aspects of Stress Corrosion Cracking," Staehle, R.W et al,

Eds., National Association of Corrosion Engineers, Houston, Tex., 1969, p 104

'WeUs, A.A., Metals and Materials, Vol 3, 1969, p 173

Trang 22

A A Sheinker,^ and J D Wood^

Stress Corrosion C r a c k i n g of a

High Strength Steel

REFERENCE: Sheinker, A.A and Wood, J.D., "Stress Corrosion Cracking of a

High Strength Steel," Stress Corrosion Cracking of Metals-A State of the Art,

ASTM STP 518, American Society for Testing and Materials, 1972, pp 16-38

ABSTRACT: Stress corrosion crack growth rates (dA/dT) as a function of stress

intensity factor (K) were determined over a wide range of electrode potentials for

AISI 4340 steel (200-ksi yield strength level) in deaerated 3.5 percent sodium

chloride solution buffered to pH 3.8 Particular emphasis was placed on conducting

the stress corrosion tests under well defined electrochemical and mechanical

conditions At intermediate K levels, dA/dT was essentially independent of K,

suggesting that crack growth is limited by mass transport Crack growth is

apparently dominated by localized mechanical rupturing at high K levels where

dA/dT increased rapidly with increasing K Except at a very cathodic potential,

dA/dT at intermediate K levels was also independent of potential, implying that

the electrochemical conditions at the tip of the stress corrosion crack are not the

same as those outside the crack The tendency for the stress corrosion cracks to

branch was found to be electrochemically, as well as mechanically, controlled

KEY WORDS: stress corrosion cracking, fracture properties, mechanical properties,

high strength steels, electrochemistry, hydrogen embrittlement, failure, crack

propagation, crack initiation, stress analysis, plastic deformation

Definition of Stress Corrosion Cracking

Stress corrosion cracking is the failure of a metal resulting from the

conjoint action of stress and chemical attack There are two different types of

mechanisms by which stress corrosion cracking is beheved to occur: active

path corrosion and hydrogen embrittlement In the active path corrosion type,

cracking is caused by localized corrosion at the crack tip, and proceeds along

a path which is electrochemically active with respect to the surrounding metal

In the hydrogen embrittlement type of mechanism, cracking results from the

entry of hydrogen into the metal, which reduces its ability to deform

plastically

Department of Metallurgy and Materials Science, Lehigh University, Bethlehem, Pa

18015

16

Trang 23

SHEINKER AND WOOD ON A HIGH STRENGTH STEEL 17 Since hydrogen embrittlement is not a corrosion process in the classical

sense, cracking occurring by this mechanism is sometimes excluded from the

term "stress corrosion cracking." However, since it is often not clear which of

the two types of mechanisms is responsible for cracking, it is convenient and

practical to use the generic term "stress corrosion cracking" to refer to

cracking occurring by either type of mechanism This usage is adopted in this

paper

In order for stress corrosion cracking to occur, the following conditions are

necessary: (1) a susceptible metal, (2) a specific environment, and (3) a tensile

stress Metal susceptibility and environment specificity depend on the

partic-ular metal-environment couple A metal may be susceptible to stress corrosion

cracking in only a few specific environments, and conversely, a particular

environment may induce cracking in only certain metals The tensile stress

usually must exceed a certain level, depending on the particular

metal-environment couple, to produce stress corrosion cracking

In general, the sequence of events leading to failure of a metal by stress

corrosion cracking begins with localized chemical attack of the metal surface

A crack then initiates at a sharp intrusion produced by localized attack, and

grows slowly When the crack reaches a size at which the metal can no longer

support the load, rapid fracture occurs If a crack like flaw is already present

in the metal surface, localized attack is unnecessary, and slow crack growth

proceeds from the flaw, sometimes after a period of incubation

The process of stress corrosion cracking involves a complex interaction of

metallurgical, chemical, and mechanical factors Since these three factors

correspond to the three conditions necessary to produce stress corrosion

cracking, it is evident that the role of each factor must be understood to

completely understand this phenomenon Much emphasis has been placed on

the metallurgical aspect of stress corrosion cracking, but considerable progress

is being made in understanding the chemical aspect by the application of the

principles of electrochemistry, and in understanding the mechanical aspect by

the apphcation of the concepts of fracture mechanics

Use of Electrochemistry to Study Stress Corrosion Cracking

It is well known that corrosion of metals is an electrochemical process, and

it is also now well established that stress corrosion cracking of metals in

aqueous solutions is governed, at least to some extent, by electrochemical

reactions [1] Previous studies of the effects of electrochemical variables on

stress corrosion cracking have been mainly devoted to the determination of

the relation between electrode potential and the time-to-failure of stressed

specimens for various metal-environment systems For steels, this so-called

"electrochemical polarization method" has been used to determine whether

the mechanism of stress corrosion cracking is one of active path corrosion or

hydrogen embrittlement [2] If anodic polarization decreases the failure time

with respect to that at the "open-circuit" or "rest" potential, then the

Trang 24

mechanism is said to be of the active path type Conversely, if cathodic

polarization decreases the failure time with respect to that at the rest

potential, then the mechanism is said to be one of hydrogen embrittlement

Most of these studies, however, have been deficient in that they were not

conducted under well defined electrochemical conditions, including solution

pH and dissolved ion and oxygen contents These conditions probably varied

during the course of the experiments In some, the pH of the corrosive

solution was adjusted by the addition of simple acids and bases which have

low buffer capacities, that is, ability of the solution to resist changes in

hydrogen ion concentration \3] This is an important consideration, as it has

been observed that the pH of the solution within a stress corrosion crack is

markedly different from that of the external bulk solution [4-6]

Application of Fracture Mechanics to Stress Corrosion Cracking

The traditional method of evaluating the stress corrosion susceptibility of a

metal is the determination of the time required to produce failure

(time-to-failure) in smooth specimens tested at different gross stress levels in the

appropriate corrosive environment However, the gross stress is not an accurate

measure of the mechanical driving force for crack propagation, since it does

not relate to the tip of the crack where the stress corrosion process operates

In addition, the time-to-failure of smooth specimens incorporates the time

required for both crack initiation and slow crack growth so that the separate

effect of the environment on each of these stages cannot be ascertained

An accurate quantitative description of the mechanical driving force for

stress corrosion crack propagation is provided by the stress intensity factor K

from fracture mechanics [7-9] K is a measure of the intensity or magnitude

of the stresses near the tip of a crack in a solid body [10] It is a function of

the loading and the configuration of the body, including the crack size The

units for K are those of (force/area) x (length)''^, and in the English system

the units are psi TlrT or ksi Tin

The crack tip stress analysis is based on the assumption of hmited

plasticity, so that the use of K is vaUd only when plasticity at the crack tip is

limited This is true when the state of stress at the crack tip is essentially that

of plane strain, which requires minimum values of crack size and section

thickness [1]]

In order to utiUze fracture mechanics concepts in stress corrosion studies,

precracked specimens, into which cracks have been dehberately introduced,

must be employed Besides enabling the use of K to characterize the crack

driving force, the use of precracked specimens offers several other advantages

over smooth specimens Precracking eliminates the localized attack stage,

which is highly variable and often not representative of material behavior in

service It also avoids variations in the metal surface, to which the localized

attack stage is very sensitive In addition, the use of precracked specimens

facilitates measurement of the slow crack growth rate

Trang 25

SHEINKER AND WOOD ON A HIGH STRENGTH STEEL 19 Much of the stress corrosion research utilizing the fracture mechanics

approach has been concerned with the determination of the relation between

the initially applied stress intensity factor /ij and the time-to-failure of

precracked specimens for various metal-environment systems [7-9] The

time-to-failure usually increases as ATj is decreased from the level required for

the onset of rapid fracture (designated K^^, the "fracture toughness") to a

threshold level (designated A'jjj,^) below which stress corrosion cracking does

not occur (The fracture toughness and stress corrosion cracking threshold are

designated A^j;, and A^isc^ , respectively, only when the state of stress at the

crack tip is essentially that of plane strain.) However, neither K-^ nor the

failure time are fundamental parameters For most specimen configurations,

under constant load, K increases as the crack propagates through the

specimen, with the relation between K and crack length dependent upon the

particular configuration The time-to-failure depends on (1) the size and

configuration of the specimen, (2) the length of the incubation period which

may precede the commencement of slow crack growth, and which may be

/^-dependent, (3) the relation between crack growth rate (dA/dT) and K, and

(4) the fracture toughness of the metal, which determines how far the crack

will grow before rapid fracture occurs

Since K is an accurate measure of the mechanical driving force for stress

corrosion crack propagation, a unique relationship between dA/dT and the

instantaneous K level should exist for a given metal-environment system,

provided the metallurgical and environmental conditions are held constant

Relatively little is known about the relation between dA/dT and K for various

metal-environmental systems For high strength steels-jn distilled or salt water,

linear [12-14] and logarithmic [15,16] relationships have been reported, and

it has also been observed that dA/dT is independent of K over a wide range of

K levels [13,17] However, none of these reported relationships was obtained

under well defined electrochemical conditions In addition, many of these

relationships may be misleading because, either they were obtained over a

narrow range of K relative to the range Ki^^^ to /Tj^,, or they are based on

relatively few data points

Stress Corrosion Cracking of High Strength Steels

The term "high strength steels" here refers to martensitic,

precipitation-hardenable, and maragjng steels, which have yield strengths in excess-of about

150 ksi These steels are susceptible to stress corrosion cracking in a wide

variety of both aqueous and nonaqueous solutions [2] The susceptibility of

these steels generally increases with increasing yield strength

Stress corrosion cracks in high strength steels usually initiate at the bases of

sharp corrosion pits which invariably form at the sites of nonmetallic

inclusions in the surface of the metal [2] The path of crack propagation is

usually intergranular, but is sometimes transgranular, with respect to the prior

austenite grain boundaries

Trang 26

There has been no general agreement on whether the mechanism of stress

corrosion cracking in high strength steels is one of active path corrosion or

hydrogen embrittlement, or possibly a combination of both types of

mech-anisms [2] Use of the electrochemical polarization method has indicated that

either type of mechanism may be operative, depending on the particular steel

and the nature of the environment However, in recent years, the following

evidence has been produced which indicates that hydrogen embrittlement may

be the major mechanism, at least in martensitic high strength steels:

1 The electrochemical conditions (solution pH and electrode potential) at

the tip of a stress corrosion crack in a high strength steel are apparently

favorable for hydrogen liberation regardless of the external electrochemical

conditions [6]

2 Hydrogen can permeate through steel under anodic as well as cathodic

polarization [18]

3 Hydrogen permeation correlates with stress corrosion susceptibility [18]

4 Martensitic structures absorb atomic hydrogen at an increasing rate with

increasing tensile stress [19]

5 Atomic hydrogen diffuses to and concentrates in the region of highest

tensile stress [19]

Purpose of This Investigation

The purpose of this investigation was to gain a better understanding of how

the kinetics of stress corrosion cracking in high strength steels are controlled

by electrochemical reactions The objective was the determination of the

effect of electrode potential on the slow crack growth rate as a function of

stress intensity factor Particular emphasis was placed on conducting the stress

corrosion tests under well defined electrochemical and mechanical conditions

Experimental Procedure

Material and Specimens

The material selected for this study was AISI 4340 steel heat treated to a

yield strength of about 200 ksi This material was obtained in the form of a

1-in thick hot rolled plate The chemical composition of this steel is

-Composition of AISI 4340 steel,

P 0.005

S 0.008

Si 0.35

Ni 1.69

weight

Cr 0.88

' percent

Mo 0.21

Al 0.03

In order to utilize the stress intensity factor K to characterize the

mechanical driving force for stress corrosion crack propagation, precracked

Trang 27

SHEINKER AND WOOD ON A HIGH STRENGTH STEEL 21

cantilever bend specimens were employed In this type of precracked

speci-men, under constant load, K increases rapidly with increasing crack length,

thereby providing a large variation in K within a single specimen

The specimen dimensions are shown in Fig 1 The thickness (0.5 in.) was

sufficient to ensure a plane strain state of stress at the crack tip at all K levels

attained The specimen depth (1.5 in.) was large enough to provide a

reasonable amount of slow crack growth prior to rapid fracture

-FIG 1 -Stress corrosion test specimen

The circular notch in the upper edge of the specimen was used to provide

knife edges for attaching a crack opening displacement gage The straight

notch below the circular notch was used to promote initiation of the fatigued

precrack The grooves in the sides of the specimen were employed to prevent

crack branching

The specimen blanks were cut from the 1-in thick plate such that the

applied stress direction and the crack growth direction were in the RW

orientation, that is, wath the specimen length (applied stress direction) parallel

to the plate rolling direction and the specimen depth (crack growth direction)

parallel to the width of the plate Equal amounts of material were machined

Trang 28

from both sides of the blanks so that the resulting specimen thickness

corresponded to the middle of the plate thickness

The specimens were heat treated as follows:

1 Normalized at 1700 F for 1 h in an inert atmosphere

2 Austenitized at 1550 F for 1 h in an inert atmosphere

3 Quenched in still oil at room temperature

4 Immersed in liquid nitrogen for 30 min

5 Tempered at 750 F for 1 h in air, and air cooled

The hardness of the heat treated specimens was Rockwell C 48 ± 0.5

Following heat treatment, the specimens were ground to final dimensions

The straight notch was cut by electrical discharge machining The specimens

were precracked by cychc loading, with a maximum K level in the final

increment of fatigue crack growth of 15 ksi/Iri The precrack length was 0.5

in including the circular and straight notches The side grooves were machined

after precracking The specimens were stored in a dessicator until ready for

stress corrosion testing

Prior to stress corrosion testing, the specimen was vviped clean with acetone

followed by methanol In order to restrict access of the corrodent to the

mouth of the crack, the surface of the specimen to be exposed to the

corrodent, with the exception of the circular and straight notches, was then

masked with an inert polymeric coating

Stress Corrosion Tests

A series of stress corrosion tests was conducted at electrode potentials

ranging from +240 to —960 mV on the standard hydrogen electrode (SHE)

scale These tests were conducted under carefully controlled electrochemical

conditions In each test, the corrodent was deaerated and the electrode

potential of the specimen was held constant All the tests were conducted at

room temperature in an air conditioned laboratory

The corrodent employed in the stress corrosion tests was an aqueous

solution of 3.5 percent (0.6M) sodium chloride buffered to a pH of 3.8 by

the addition of 1.02 percent (0.05M) potassium biphthalate (KHC8H4O4)

This pH level was chosen because it has been found to be the pH at the tip of

a stress corrosion crack in several high strength steels, including AISI 4340, in

the absence of an applied potential, regardless of the pH of the bulk solution

outside the crack [4,5] The water used to prepare the corrosive solution was

deionized water having a specific conductance no greater than 2 x 10"* (ohm

• cm)''

The precracked section of the specimen was immersed in the corrodent

using an enclosed Plexiglas chamber, as shown in Fig 2 Holes were provided

in the top and sides of this chamber for insertion of various instrument

probes The specimen was sealed to the sides of the chamber with the same

polymeric material used to mask the specimen The corrodent was fed to the

Trang 29

A—Reference electrode

B—Gas vent tube

C—Luggjn capillary probe

D—Thermometer

E—Gas dispersion tubes F—Platinum auxiliary electrodes G—Crack opening displacement gage H—Dissolved oxygen probe

FIG 2—Stress corrosion test setup with instrument probes

environment chamber by gravity flow from a 5-gal supply tank, and was

allowed to flow through the chamber at a rate of about 30 ml/min to prevent

accumulation of corrosion products in the chamber The corrodent flow rate

through the chamber was controlled by means of inlet and outlet stopcocks

The pH of the effluent solution was measured periodically with a pH meter,

and remained constant throughout each test

The corrodent was deaerated by bubbling purified nitrogen gas through the

solution in the supply tank This gas was also bubbled through the solution in

the environmnet chamber by means of two fritted glass gas dispersion tubes

immersed in the solution on either side of the specimen, as shown in Fig 2

This arrangement also served to agitate the solution around the specimen The

dissolved oxygen content of the corrodent in the chamber was monitored by

means of a dissolved oxygen meter with a polarographic probe immersed in

the solution (Fig 2) The dissolved oxygen content in all of these tests was

Trang 30

The electrode potential of the specimen was controlled by means of an

electronic potentiostat with two platinum auxiliary electrodes and a saturated

calomel reference electrode The two auxiHary electrodes were immersed in

the corrodent on either side of the specimen, but the reference electrode was

immersed in the corrodent in a separate cell beside the chamber (Fig 2) This

cell was connected to the chamber by a Luggin capillary probe to form a salt

bridge To eliminate the voltage drop through the solution from the potential

measurement, the tip of this probe was positioned about 1 mm from the side

of the specimen and midway down the depth of the specimen, at the

precracked section

The specimen was loaded as a cantilever beam by clamping one end to a

rigid test stand, and the other end to a lever arm, as shown in Fig 3 Disk

shaped weights were then suspended from the other end of the lever arm to

obtain the desired bending moment The load was applied after the specimen

was immersed so the corrodent would be immediately drawn into the crack

Each test was conducted at a constant load with an initial K level of about 25

A-Environment chamber B-Test stand

C—Lever arm D-Weight

FIG ^-Arrangement for loading test specimen

Trang 31

SHEINKER AND WOOD ON A HIGH STRENGTH STEEL 25 ksi/in., since crack growth did not commence in a reasonable length of time

(several hours) at lower K levels (The K^^^^ for this material in sodium

chloride solutions is about 10 to 15 ksi /iii [75].)

Stress corrosion crack growth was monitored by means of a crack opening

displacement gage of the clip-on type used in fracture toughness testing [20]

This gage was attached to the upper edge of the test specimen on the knife

edges formed by the circular notch, as depicted in Fig 4 As the crack

propagated through the specimen, the crack opening displacement increased as

a function of the crack length The relationship among applied moment, crack

length, and crack opening displacement was previously determined by

per-forming a compliance calibration on a test specimen This relationship is

plotted in Fig 5, in which the points represent the actual data from the

compliance calibration, and the curve represents a least-squares fit to this data

for the relation

L" M ] • i t ] -h D

where

V - crack opening displacement, in.,

B = gross specimen thickness, in

' ;**' v ^ "^

• i

FIG A-Crack opening displacement gage attached to test specimen

Trang 32

26 STRESS CORROSION C R A C K I N G OF M E T A L S

E = Young's modulus, psi,

W = specimen depth, in.,

M = applied moment, lb ° in.,

RELATIVE CRACK LENGTH, A/W

FIG 5-Compliance relationship for side grooved cantilever bend specimens

The electrical response of the gage was continuously recorded by means of

a strip chart recorder Data points were taken from the recorder chart at equal

increments of gage response and converted to values of crack opening

displacement by means of a prior calibration of the gage The displacement

values were then converted to crack lengths by means of the compliance

relationship The crack growth rate was determined by calculating the slope of

the crack length-time curve at each data point

Trang 33

SHEINKER AND WOOD ON A HIGH STRENGTH STEEL 27 The Stress intensity factor corresponding to each data point was calculated

by means of the equation

2

where

M = applied moment, lb • in.,

A = total crack length, including circular and straight notches, in.,

W = specimen depth, in.,

B = gross specimen thickness, in., and

5yv = net specimen thickness between side grooves, in

This equation is for single edge cracked plate specimens subjected to pure

bending [11], modified by the factor {B/Bj^Y'^ to include the effect of the

side grooves [21]

All the calculations were performed by a high speed digital computer

Graphs of crack length (.4) as a function of time (7), and crack growth rate

(dA/dT) as a function of stress intensity factor (K), for each stress corrosion

test, were plotted by an automatic plotter connected to the computer

Results and Discussion

General Shape of dA/dT-K Curves

Stress corrosion crack growth proceeded from the precrack length of 0.5

in to a final crack length of 0.8 to 1.0 in., at which the specimens failed by

rapid fracture The stress corrosion fracture surface was very rough compared

with the fatigue and rapid fracture surfaces, as shown in Fig 6 Typical crack

length-time plots are shown in Figs 7 and 8 The plot in Fig 8 indicates an

incubation period of about 30 min preceding the commencement of stress

corrosion crack growth An incubation period was observed in some of the

tests, but did not vary regularly with electrode potential

A typical plot of dA/dT versus K is presented in Fig 9 All of the

dA/dT-K plots had this same general shape regardless of the applied potential

As K increases, dA/dT passes through several stages At low K levels, dA/dT

increases rapidly with increasing K (Stage I) At intermediate K levels, dA/dT

is either independent of K, or only shghtly A'-dependent (Stage II) At higher

K levels, dA/dT again increases rapidly with increasing K (Stage III) This

Trang 34

A—Circular notch B-Straight notch C—Fatigue precrack

D—Stress corrosion crack E—Rapid fracture F—Side grooves

FIG b-Typical fracture surface of stress corrosion test specimen

Stage is followed by the onset of rapid fracture, at which dAjdT rises

abruptly

Stage I of the dA/dT-K curves probably has little physical significance since

it is associated with the commencement of stress corrosion crack growth from

the existing fatigue precrack The maximum K level in the final increment of

fatigue precracking (15 ksi /In.) was kept well below the initial K level in the

stress corrosion tests (25 ksi Im.) so that the initial stress corrosion crack

growth rate would not be affected by the residual compressive stresses in the

plastic zone at the tip of the precrack [22] However, since the material in

Trang 35

IQ.OQD

9.0QD -•

4,000 -I 1 1 1 1 1 1 1 1 1—

O.OOO Z.OOO 4.000 6.000 8.0O0 lO.OOO

TIME, T, MINUTES XIO

FIG 1-Typical crack length-time plot for stress corrosion test with no incubation

period

this plastic zone has undergone cyclic plastic deformation, it is not

representa-tive of the material in the bulk of the specimen This could affect the initial

crack growth rate as well as the Incubation period preceding the

commence-ment of stress corrosion crack growth

In addition, the commencement of stress corrosion crack growth from the

precrack may not be uniform along the precrack front, so that the crack

extends farther at some points along its front than at others The rate of

crack opening displacement, and thus the apparent crack growth rate, would

then increase as the crack front became uniform Thus, the apparent

/k-dependence of dA/dT in Stage 1 may simply be a manifestation of a

transition from an unevenly extending crack front at the precrack tip, to a

uniformly extending crack front

Trang 36

3 0 STRESS CORROSION CRACKING OF METALS

The occurrence of the crack growth rate plateau at intermediate K levels

(Stage II) suggests that crack growth in this stage may be limited either by

mass transport or by the kinetics of a reaction between the metal and the

corrodent From a purely mechanical standpoint, the crack growth rate would

be expected to increase as the stress intensity at the crack tip is increased, as

is the case in fatigue crack propagation [22] The mass transport involved in

limiting crack growth could be either the diffusion of hydrogen into the metal

at the crack tip, or the migration of corrosive reactants down the length of

the crack (The impervious coating masking the sides of the specimen

prevented direct access of the corrodent to the crack tip.) Alternatively, the

crack growth rate may be controlled by the rate of some reaction at the crack

tip, such as the adsorption of hydrogen on the metal surface

Trang 37

ao.ooo x o o o 40.000 so.ooo eo.ooo TO.OOO

STRESS INTENSITY FRCTOR K KSI\/irn

FIG 9~Typical dA/dt-K plot

There is some evidence available which indicates that the crack growth rate

in Stage II is controlled by mass transport rather than by reaction kinetics In

preliminary tests on two specimens with no side grooves and no coating

masking the sides, polarized to -760 and -960 mV (versus SHE) in unbuffered

3.5 percent NaCl solution, the crack propagated more rapidly along the sides

Trang 38

of the specimens than in the interior, as shown in Fig 10 The curvature of

the stress corrosion crack front at the onset of rapid fracture (indicated by

the arrows in Fig 10) suggests that crack growth is mass transport-controlled

FIG IQ—Fracture surface of specimen with no side grooves and no

coating masking the sides, polarized to -960 mV (versus SHE) in

unbuffered 3.5 percent NaQ solution

Mass transport-hmited crack growth is apparently overridden by the

mechanical factor in Stage III, where dA/dT increases rapidly with increasing

K Crack growth in this stage appears to be dominated by localized bursts of

brittle rupture, as it is accompanied by audible acoustic emissions Some of

these bursts are quite large, as evidenced by the peaks preceding the onset of

rapid fracture in Fig 9 These bursts probably resulted from the local fracture

toughness being exceeded at isolated locations along the path of the stress

corrosion crack

Trang 39

Secondary variations in dAjdT appeared in all of the dA/dT-K plots,

indicating that crack growth proceeds irregularly These variations may be due

either to the inherently discontinuous nature of stress corrosion crack growth,

or to secondary branching The formation of nonpropagating branch cracks

would temporarily increase the rate of crack opening displacement, and thus

result in transitory increases in the apparent crack growth rate

Reproducibility of dA/dT-K Curves

The stress corrosion test results are summarized in Table 2 The average

dA/dT in the range A^ = 30 to 40 ksi Jm is representative of the crack

growth rate in Stage II which consists of a growth rate plateau The K level at

failure (rapid fracture) is equivalent to the fracture toughness of the metal,

though the latter is measured more accurately under increasing load in

accordance with the ASTM Test for Plane-Strain Fracture Toughness of

Metallic Materials (E399-70T)

TABLE 2~Stress corrosion test results

mV t240 -110 -340 -340

- 5 6 0 -560 -760

- 7 6 0

Average Current

mA ISA"

3.0A 0.33A 0.50A 1.3C 1.3C 3.0C 3.7C

Avg dA/dT

K=30ioK

in./min 0.0042 0.0055 0.0039 0.0037 0.0042 0.0032 0.011 0.0069

"A indicates current was anodic;

C indicates current was cathodic

*Apparent value

The degree of reproducibility of the dA/dT-K curves can be determined by

comparing the results for specimens tested at the same potential With respect

to the average dA/dT in the range K = 30 Xo 40 ksi Jm., the reproducibility

is excellent for the two specimens tested at -340 mV (versus SHE), good for

the two specimens tested at —560 mV, and only fair for the two specimens

tested at -760 mV

The reproducibility is generally poor for Stage III, partly because of

the discontinuous nature of the brittle rupture bursts which characterize

this stage, and partly because of the variation in the K level at failure

Trang 40

With the exception of Specimen 25, the variation in K level at failure,

48 to 57 ksi /In., is about the same as that normally found in fracture

toughness values In Specimen 25, at /T ~ 43 ksi Jm., the crack began

to branch on one side of the crack front, as indicated by the arrow in

Fig 11 This behavior reduced the actual K at the crack tip, so that the

FIG 11 -Fracture surface of Specimen 23

amount of slow crack growth was extended, and rapid fracture occurred

at a higher apparent K level (73 ksi Jm.)

Effect of Potential

The rest potential of the specimens in this corrodent was about —340 mV

(versus SHE) (More positive potentials were anodic, whereas more negative

potentials were cathodic.)

Tiêu đề	Stress Corrosion Cracking of Metals
Tác giả	H. Lee Craig, Jr.
Trường học	University of Washington
Chuyên ngành	Metallurgy
Thể loại	Báo cáo kỹ thuật đặc biệt
Năm xuất bản	1983
Thành phố	Philadelphia

Định dạng
Số trang	173
Dung lượng	3,19 MB